The gender system of Tashlhiyt: Using supervised learning to predict noun gender
John Alderete, Abdelkrim Jebbour, Kaye Holubowsky, Piyush Agarwal
January 2025 

This article develops a comprehensive account of the system of gender assignment in Tashlhiyt and the first quantitative account of gender in an Amazigh language. From a corpus of 1914 noun paradigms, we establish the number of genders, gender morphology, and the masculine default in Tashlhiyt within contemporary theory of gender assignment. While much prior work focused on gender in wordforms, we explore form and meaning attributes of word lemmas and use them to predict Tashlhiyt gender with a set of computational classifiers. The resulting quantitative analysis reveals important roles for previously unacknowledged morphological and phonological attributes, and it also presents a broader argument for using machine learning techniques in the linguistic analysis of gender assignment. All data and models are open access.

Format:	[ pdf ]
Reference:	lingbuzz/008744 (please use that when you cite this article)
Published in:	under review
keywords:	gender assignment, typology, tashlhiyt, amazigh, machine learning, classification, supervised learning, semantics, morphology, phonology
Downloaded:	208 times

 

[ edit this article | back to article list ]